The Dot-Com Bubble: What Happened and Why

By Victor Hoang · Updated Oct 23, 2026

In 1995 the network that universities had used for a decade was suddenly open to everyone. No one knew which Internet businesses would earn money. A stock market eager to find out was willing to pay for the guess.

The dot-com bubble was the run-up and collapse of Internet-company stock prices in the late 1990s. It followed the commercialization of the Internet in 1995. The Nasdaq Composite closed at a record 5,048.62 on 10 March 2000. By 9 October 2002 it closed at 1,114.11, about 78% lower. The throughline: a real technology opened up, and a story about it ran ahead of revenue. Cheap public capital funded the story, and builders laid more network than demand could fill.

This article follows how that guess grew into a bubble, how it broke, and what survived. It is one chapter of the history of the Internet, and the short version of the same dates sits in the Internet timeline.

A quick definition. A "bubble" here means prices that rose well beyond what the businesses behind them earned, until a reversal. Whether any one price was too high is a judgment. The 2000 to 2002 fall is a measured fact.

Timeline at a Glance

The Precursor Manias (1720-1929)

Earlier crowds also priced a story, not a profit

Start with a precursor, not the concept. Share-price manias are far older than the Internet, and the dot-com bubble was not the first.

In London in 1720, the South Sea Company's plan to take over government debt in exchange for its own shares drew a flood of buyers. A member of the House of Commons, Archibald Hutcheson, printed a folio pamphlet, now held by Sir John Soane's Museum, with a preface dated 31 March 1720. It worked out how much new subscribers would lose, and the museum notes that prices rose sharply from January to August before the market collapsed in September.

The 1920s give a second picture, this time with leverage. The Federal Reserve's history of the 1929 crash says buyers could put down a fraction of the price, typically 10 percent, and borrow the rest. The Dow rose from 63 in August 1921 to 381 in September 1929, a six-fold increase, fell nearly 13 percent on Black Monday, 28 October 1929, and did not regain its 1929 high until 23 November 1954.

Neither episode involved the Internet. What they supply is a pattern that the 1990s would repeat in a new setting: a plausible story, easy credit and a crowd.

For teams today: ask what a price or a valuation assumes, then ask who has to be right for it to hold. A story that needs everyone to keep buying is a warning, whatever the technology.

The Open Door (1991-1995)

NSFNET steps aside and private carriers take the traffic

The turning point is a change in who ran the pipes. The National Science Foundation's backbone had grown from about 2,000 connected computers in 1986 to more than 2 million by 1993, but its Acceptable Use Policy kept the backbone for research and education. The Internet Society's history calls the resulting push toward private, competitive long-haul networks "predictable (and intended)" and names PSI, UUNET and ANS CO+RE as early examples. NSF defunded the backbone in April 1995, and the money went to regional networks to buy national connectivity from private providers.

Picture a university lab in 1994. The network was fast, free to the department and unknown to almost everyone outside it. A year later the same wire sat inside a commercial market, with companies selling access and a browser on the screen. The NSF itself says it shut NSFNET down in 1995 because of the rapid expansion of commercial Internet services. The interface that made the network usable had come from a university too: the Computer History Museum's browser exhibit describes how Mosaic's success caught Silicon Valley's attention and led Jim Clark, founder of Silicon Graphics, to recruit Marc Andreessen in 1994.

What this solved was access: anyone could now buy a connection. What it created was a question of money. No one knew which of the new Internet businesses would earn it. For the web side of the same story, see the history of HTML and web standards and Internet vs World Wide Web.

For teams today: when an infrastructure opens to everyone, the first question is access and the second is business model. Treat the second as unanswered until revenue answers it.

The Netscape Moment (1995-1996)

A browser company becomes a public company in a day

The turning point is a stock listing. Netscape was incorporated in Delaware in April 1994 and, according to its own later SEC filing, began shipping products in December 1994. It made its initial public offering on 8 August 1995, with trading beginning on 9 August. The filing gives the offer price as $14.00 per share after a two-for-one split in January 1996, which is $28 on the original shares. A company about 16 months old was suddenly a public one.

The Computer History Museum says the IPO was a major success that left Netscape at the peak of its popularity and helped start the dot-com boom. Its model, free software for individuals and site licenses for companies, helped turn the Internet from a mostly academic network into a big commercial market. The same exhibit notes that Microsoft then built Internet Explorer on a licensed version of Mosaic and bundled it with Windows 95, and that Internet Explorer passed Netscape by 1998.

Law and money moved the same year. The Telecommunications Act of 1996, signed on 8 February 1996, describes itself as "an Act to promote competition and reduce regulation" in telecommunications and to encourage new technologies. The Richmond Fed would later say that regulatory and technological change set off a boom in telecommunications from 1997.

Then came the warning. On 5 December 1996, at the American Enterprise Institute's Francis Boyer lecture, Federal Reserve Chairman Alan Greenspan asked how we know "when irrational exuberance has unduly escalated asset values". He used Japan's recent experience as the example. The Nasdaq peaked more than three years after that speech.

What this era solved was money for the new web: a public market would fund companies that had not earned a profit. What it created was a template. Raise capital on a story, list the shares and grow first.

For teams today: a hot listing tells you what the market will pay on one day, not what the business earns. Judge a vendor or a competitor by its revenue and retention, not its funding headline.

The Dot-Com Boom (1997-March 2000)

Companies spend on growth and networks outrun demand

The turning point is a belief that the spending would pay back. In his first letter to shareholders, Jeff Bezos wrote that fundamental success would be measured by the "shareholder value we create over the long term". The same letter reports Amazon's 1997 revenue as $147.8 million, up 838% from $15.7 million, and says "this is Day 1 for the Internet". Growth was real. The question was what it cost.

Look at a warehouse in Oakland in 1999. Webvan sold same-day grocery delivery in 30-minute windows from its own distribution center. Its annual report for 1999 shows the IPO on 4 November 1999 at $15.00 a share, with 28.75 million shares sold and about $431.3 million raised, a 1999 net loss of about $144.6 million, and a plan to have centers in 15 markets by the end of 2001. The plan assumed demand would arrive on schedule.

The network builders made the same bet with steel and glass. The Dallas Fed reports that telecom firms laid $90 billion of fiber-optic cable between 1997 and 2001, about 39 million miles, and that new multiplexing technology let more data run over the same fiber. The same Dallas Fed piece says demand never came in at forecasted levels. The New York Fed's Jonathan McCarthy found that investment in communications had an outsized effect on both the end of the boom and the start of the 2001 recession, driven by overly rosy expectations about profits from new technologies.

Deals grew as well. In January 2000, AOL and Time Warner announced their merger, per the later AOL Time Warner annual report. A month later Pets.com, which had begun operations in February 1999, sold 7.5 million shares at $11.00 on 11 February 2000. On 10 March 2000 the Nasdaq Composite closed at 5,048.62, the peak close of the cycle.

What this era solved was visibility: a new market earned attention and capital quickly. What it created was a bill. Revenue had to catch up with spending, and for many firms it did not.

For teams today: growth funded by someone else's capital is a bet on the next round. Ask how long a plan can run if the money stops, because that is the real planning horizon.

The Dot-Com Crash (2000-2002)

Funding stops, fiber sits idle and accounts are restated

The turning point is the peak itself. From 10 March 2000 the Nasdaq fell, and by 9 October 2002 it closed at 1,114.11, about 78% below the record close. The NBER dates the US recession from March to November 2001, ending an expansion that had lasted 120 months from March 1991.

Take Pets.com as the compact case. According to its SEC filings, the company lost $146.6 million from inception to 30 September 2000. Its board adopted a plan of complete liquidation on 4 November 2000, per its later annual report, which also says the board acted after no viable offers to acquire or fund the company were received. Stockholders approved on 16 January 2001 and the dissolution took effect on 18 January 2001.

The telecom side fell harder than a single retailer could. The Dallas Fed reports that usage of the long-distance backbone fell below 3 percent in April 2001, and that sources of easy money for anything with "telecom" in the name dried up, leaving heavy debt and a rise in bankruptcies. Then came fraud. On 26 June 2002 the SEC filed a complaint in the Southern District of New York charging WorldCom with an accounting fraud totaling more than $3.8 billion, capitalizing costs that should have been expensed.

The biggest merger of the era shows how a mismatch of dates and values hurts. The AOL-Time Warner merger was announced in January 2000 and consummated on 11 January 2001. In the first quarter of 2002 the company recorded a $54.235 billion non-cash charge for goodwill impairment, substantially all of it generated in the merger, and the same filing shows a $45.538 billion write-down of goodwill and other intangible assets in 2002. The filing says the charge reflects market declines since the merger was announced.

What this era solved was the excess: weak plans ran out of cash, and the market priced risk again. What it created was a long distrust of Internet claims, and a surplus of cheap network capacity.

For teams today: price the date gap in any deal. A merger agreed at one price and closed a year later can be a different deal by the time it closes.

The Survivors (2002-present)

The network outlasts the stocks and carries a new boom

The turning point is that the Internet itself never fell. Many of the companies did. Amazon, which had reported $147.8 million in 1997 revenue, kept building under the long-term logic of its first shareholder letter. Google was incorporated in August 1998 when Andy Bechtolsheim wrote a $100,000 check to Larry Page and Sergey Brin. Both firms were products of the boom years that outlasted it. The Dallas Fed noted at the time that while billions went into long-haul networks, little was done on the last mile into homes and businesses, so the long-haul capacity stayed in the ground for whoever came next.

The step from rented servers to on-demand infrastructure is told in the history of cloud computing, and the protocols underneath trace back through the history of TCP/IP. The cycle has also come back. Our explainer on the AI bubble sets today's AI spending beside the dot-com run-up, and shows where the two differ.

For teams today: a crash tests business models, not the technology. When a market corrects, the tools that were already earning their cost tend to stay.

Key People

Key Facts

• The Nasdaq Composite closed at 5,048.62 on 10 March 2000 and at 1,114.11 on 9 October 2002.
• Telecom firms laid $90 billion of fiber between 1997 and 2001, and backbone usage fell below 3 percent in April 2001, per the Dallas Fed.
• Pets.com lost $146.6 million from inception through 30 September 2000 and began liquidating on 4 November 2000.
• The AOL-Time Warner merger, announced January 2000 and consummated 11 January 2001, led to a $54.235 billion goodwill impairment charge in the first quarter of 2002.
• The SEC's WorldCom complaint of June 2002 alleged more than $3.8 billion in improperly capitalized costs.
• The NBER dates the 2001 recession from March to November 2001.

The Pattern Behind the History

We call the sequence the Story, Capital, Overbuild Pattern. It is our synthesis of the history above, not a research finding.

1. A real technology opens. NSFNET gave way to private carriers and a browser made the network usable.
2. A story outruns revenue. Netscape's 1995 listing and Amazon's growth made "get big first" believable.
3. Cheap public capital funds the story. IPOs such as Webvan's and Pets.com's raised hundreds of millions before profits existed.
4. Builders overbuild on the forecast. Fiber went in faster than demand, and backbone usage later fell below 3 percent.
5. The bust removes the weak and leaves the network. Nasdaq fell about 78%, and the survivors used what had been built.

What It Teaches Teams Today

• Separate the technology from the price. The Internet kept growing while many valuations fell.
• Watch the funding clock. A plan that needs the next round has a horizon set by investors, not customers.
• Read the denominators. A fiber network built for forecast demand sat nearly idle when demand came in lower.
• Check the accounting behind a growth story. WorldCom's capitalized costs flattered earnings until the SEC sued.
• Price the gap between announce and close. AOL and Time Warner agreed in January 2000 and closed in January 2001.
• Keep the tools that earn their cost. The same discipline applies to AI spending now.

The dot-com bubble is easy to remember as a list of failed names. The record shows a harder lesson: a true technology can carry a false price, and both stories end up in the same history. For the surrounding facts, read who invented the Internet and the history of the Internet.